**2.4 Statistical analysis of data**

The lethal concentrations (LC50 and LC95 i.e., concentrations capable of killing 50 and 95% of the population treated) of the products tested against *A. pomi* were determined by Probit analysis [55] using SPSS version 21 (IBM Corp. Released 2015. Armonk, NY: IBM Corp.). The values were considered significantly different, when their 95% confidence intervals did not overlap. To identify the toxicity caused exclusively by OMW, the mortalities recorded with each concentration were corrected with Abbott formula [56]. The comparison of the lethal effects of OMW and the pesticide on aphid stages was performed by two-factor variance analysis (Stages\* products) followed by the Tukey test (HSD) at 5% as post-hoc for the multiple comparison of means. The homogeneity and normality of the variance of the dependent variables were verified by Levene and Shapiro–Wilk tests, respectively; the dataset was transformed into arsin √ (percentage) before analysis of variance according to Sokal and Rohlf [57]. Linear models relating stage mortality to OMW concentration have been established; their choice was made based on the low values of the Akaike Information Criterion (AIC) and standard errors as well as on the high values of R2 and their significance by the analysis of variance (F). To classify

the different stages of the aphid according to their responses to OMW, a hierarchical classification was carried out on the LC50 and LC95 using Statistica version 7 software (Statsoft Inc. USA). The data are summarized as graphs or tables.
